AT&T Wi-fi site visitors shaping apparently making some web sites unusable
After I’m dwelling in my RV, wi-fi service suppliers are my main supply of connectivity. So when both AT&T or Verizon make main adjustments, I take discover.
I lately observed that a number of web sites are fairly gradual when shopping with my AT&T marketing strategy, listed in AT&T Premier (enterprise account administration UI) as “Wi-fi Broadband Extremely for Router or Hotspt (sic)”. That is an “limitless” 100Mbit plan with 50GB for Enterprise Quick Monitor (prioritized) information. Being that I used to be far beneath the 50GB of month-to-month Quick Monitor information, my information ought to have had high precedence, so I turned suspicious. To be trustworthy, I’ve by no means observed a discernible distinction between Quick Monitor and non-Quick Monitor information charges. That is all to say that I’ve no cause to imagine that I’m being deprioritized on account of utilization.
Naturally, the primary hing I did was conduct a pace check. I already knew from earlier expertise that for some cause, AT&T site visitors to fast.com is throttled. Why AT&T desires bandwidth to look decrease than actuality is a thriller to me, however I digress. Linode.com has pace assessments that AT&T has no particular remedy for, and the nearest one to me was in Fremont, CA.
The speedtest revealed 21Mbps down and 4.5Mbps up – fairly cheap in a comparatively rural space like Durango, CO. Latency was ~130ms. That pace definitely wouldn’t clarify why it took anyplace between 15 seconds and a pair of minutes to load strava.com.
So I opened up the “Community” tab in Firefox and will clearly see that dozens of assets from cloudfront.com have been taking a number of seconds to load. The issue clearly has one thing to do with Cloudfront.
Is Cloudfront having issues? That’s simple sufficient to confirm; my Sierra Wireless RV55 CAT-12 LTE-A router additionally comprises a limiteless Verizon Enterprise SIM card that I can use to conduct assessments on Cloudfront, unbiased of AT&T.
I observed that certainly one of Strava’s javascript assets clocked in at 1.68MB, making it a pleasant check topic for pace assessments (https://web-assets.strava.com/assets/federated/find-and-invite-friends/827.js). On the time of writing web-assets.strava.com resolves to dgpcy4fyk1eox.cloudfront.web, so relaxation assured, we’re coping with Cloudfront.
After switching to Verizon, I may see that Cloudfront was having no issues. Our pal 827.js downloaded in simply over 1 second at 1.4MB/s. I clearly noticed earlier within the Firefox community tab that this useful resource took practically 1 minute to load on AT&T.
Whereas
wgetshouldn’t be my goto for command line HTTP fetching, it shows switch charges in a human pleasant format by default, so I used the next as my check case:wget -O /dev/null -q --show-progress https://web-assets.strava.com/property/federated/find-and-invite-friends/827.js
So the issue isn’t Cloudfront, as a result of Verizon was quick sufficient. It wasn’t blazing quick by any means, however I additionally didn’t have to attend 2 minutes to study whether or not Strava awarded me King of the Moutain on an area path (I wasn’t).
Perhaps this can be a world AT&T drawback. That’s simple to check as effectively – my iPhone can be on the identical AT&T enterprise account as my data-only plan, so I turned on the iPhone hotspot and made it the router’s WAN system to verify we’re altering a single variable at a time. I carried out one other pace check, revealing 23Mbps down and 3Mbps up. Nothing shocking there – regular bandwidth fluctuations for a wi-fi system. How about our wget check? 1.7MB/s. The issue is clearly not all of AT&T wi-fi.
Let’s return to the unique configuration: join on to my AT&T data-only plan with my router and re-run the wget check. Perhaps I used to be imagining issues. I’m considerably shocked to see the wget check with a switch price of ~30KB/s. I imagine this guidelines out AT&T with a souring peering settlement someplace between me and Cloudfront. My cellphone site visitors to Cloudfront is unaffected; it’s solely my data-only plan that’s affected.
I now have a reasonably clear image of what’s seemingly occurring. Good old-fashion site visitors shaping. Now that the router is related on to AT&T, the true check of site visitors shaping is switch charges whereas related to a VPN. I’ll let the picture converse for itself.
As of the time of writing, I’m not sure what’s inflicting such a big slowdown. It has rendered some web sites successfully ineffective. Every little thing on this writeup signifies, to me, that AT&T is engaged in extraordinarily aggressive site visitors shaping for some plans, rendering many web sites practically unusable.
Do you could have any concepts learn how to diagnose this drawback additional? Have you learnt one of the best ways have interaction AT&T’s technical of us to take this severely? Write me at att-traffic-shaping @ this area. I’ll add updates right here if something adjustments, or I get a response on https://bizcommunity.att.com.
Replace #1 (19:03:09 UTC)
Here’s a pcap file captured with sudo tcpdump host 'dgpcy4fyk1eox.cloudfront.web' -w seize.pcap: capture.pcap
I’ve additionally tried to disable/allow IPv6 on my native machine and my VPN connection to verify for any variations. No variations have been noticed.
Replace #2 (19:43:00 UTC)
Traceroutes
Related to AT&T iPhone through Wifi hotspot
traceroute web-assets.strava.com 130 ↵
traceroute to web-assets.strava.com (99.84.208.10), 30 hops max, 60 byte packets
1 _gateway (172.20.10.1) 4.386 ms 4.280 ms 5.292 ms
2 107.243.82.3 (107.243.82.3) 139.792 ms * *
3 * * *
4 * * *
5 * * *
6 * * *
7 * * *
8 * * *
9 * * *
10 * * *
11 * * *
12 * * *
13 12.117.216.210 (12.117.216.210) 244.538 ms 369.608 ms 933.487 ms
14 * * *
15 * * *
16 * * *
17 * * *
18 * * *
19 * * *
20 * * *
21 server-99-84-208-10.iad79.r.cloudfront.web (99.84.208.10) 206.664 ms 206.643 ms 206.623 ms
Related to router on AT&T marketing strategy
traceroute web-assets.strava.com
traceroute to web-assets.strava.com (99.84.208.115), 30 hops max, 60 byte packets
1 _gateway (192.168.13.31) 61.896 ms 61.950 ms 62.064 ms
2 172.26.96.161 (172.26.96.161) 132.156 ms 132.775 ms 132.749 ms
3 107.72.231.188 (107.72.231.188) 132.743 ms 132.718 ms 132.199 ms
4 * * *
5 12.83.179.49 (12.83.179.49) 148.138 ms 148.187 ms 148.162 ms
6 slkut21crs.ip.att.web (12.122.1.186) 148.361 ms 137.643 ms 142.401 ms
7 dvmco22crs.ip.att.web (12.122.28.45) 146.657 ms 146.627 ms 149.389 ms
8 cgcil21crs.ip.att.web (12.122.28.78) 149.369 ms 149.345 ms 216.643 ms
9 12.122.28.206 (12.122.28.206) 149.302 ms 146.467 ms 149.256 ms
10 wshdc84crs.ip.att.web (12.122.135.230) 146.421 ms 149.211 ms 149.161 ms
11 wshdc406me9.ip.att.web (12.123.10.125) 143.819 ms 139.466 ms 139.378 ms
12 12.117.216.210 (12.117.216.210) 139.342 ms 139.314 ms 187.019 ms
13 * * *
14 * * *
15 * * *
16 * * *
17 * * *
18 * * *
19 * * *
20 server-99-84-208-115.iad79.r.cloudfront.web (99.84.208.115) 152.577 ms 140.190 ms 141.460 ms